Impact of protein functionality and processing on structure and texture of plant-based meat produced using extrusion and 3D printing

Abstract

Driven by demand for sustainable meat alternatives, this research investigated how protein type, functionality and extrusion processing conditions influence the structure, texture and performance of plant-based meat produced using three distinct process technologies, viz., low moisture extrusion, high moisture extrusion and additive manufacturing. These three technologies and the quality of associated products, textured vegetable protein (TVP), high moisture meat analog (HMMA) and 3D printed plant-based meat, respectively, were the focus of this study. Before evaluating the different process technologies and plant-based meat products, the functionality of various plant proteins was characterized because it strongly influenced structure development and product quality. Based on water absorption capacity (WAC) and rapid visco-analysis (RVA), soy protein isolate (SPI) and one type of soy protein concentrate (SPC1) were classified as cold swelling proteins due to their high WAC (SPI: 5.05-5.95 g/g; SPC1: 4.35 g/g) and/or low-temperature peak viscosity (30-42°C). In contrast, pea protein isolate (PPI), wheat gluten (WG), faba bean concentrate (FBC), lentil protein concentrate (LPC), and two other types of soy protein concentrate (SPC2 and SPC3) behaved as non-cold swelling ingredients, showing lower hydration and/or requiring higher temperatures to develop viscosity. Cold swelling proteins generally show high cross-linking and have greater film-forming ability, which leads to more expanded and porous plant-based meat structures, whereas non-cold swelling proteins are often associated with denser, more fibrous products. These functional differences helped explain variation in processing behavior, texture and final structure of TVP, HMMA, and 3D-printed plant-based meat. The first part of this study focused on the processing of TVP products using low-moisture extrusion and examined faba- and lentil-based formulations processed under different screw speeds and in-barrel moisture (IBM) conditions. FBC was blended with SPI, PPI, or WG, and LPC with SPI or PPI. Specific mechanical energy (SME) ranged from 595.79 to 918.86 kJ/kg in faba-based formulations and from 660.24 to 804.96 kJ/kg in lentil-based formulations. Besides protein functionality, extrusion behavior and TVP structure and texture were also affected by screw speed and IBM. Increasing IBM or lower screw speed reduced SME. SPI-containing formulations generally had the highest SME because of their high viscosity during extrusion, followed by WG and PPI. Bulk density ranged from 140.4 to 275.2 g/L and 120.7 to 266.8 g/L for FBC- and LPC-based TVP products, respectively. SPI produced more porous, lower-density structures, whereas WG and PPI produced denser, more fibrous products; bulk density was negatively correlated with water holding capacity (r = -0.93 and -0.86 for FBC- and LPC-based TVP products, respectively). These results were in alignment with the protein functionality hypothesis based on cold swelling and non-cold swelling properties. Computer vision (CV) and machine learning (ML)-based fibrosity scores showed strong positive correlations with expert visual scores for both FBC- and LPC-based TVP products (r = 0.95 and 0.96, respectively), indicating close agreement between CV+ML-based and expert assessments of fibrous structure. The second part of this study examined the processing of HMMA products using soy proteins and high moisture extrusion with a special cooling die and examined the influence of formulation and extrusion conditions on fibrous structure. A blended standard formulation (STD) consisting of SPC1, SPC2, SPI, tapioca starch, canola oil and salt was compared with a single-protein formulation (SP) consisting of SPC2, canola oil and salt, under different feed rate and barrel temperature conditions. STD showed greater hydration, lower gelation concentration, and higher viscosity development than SP, consistent with the expected functionality differences between the formulations. Increasing barrel temperature reduced SME (from 685 to 537 kJ/kg in STD and from 745 to 542 kJ/kg in SP) due to lower melt viscosity. Increasing feed rate under low-temperature conditions increased die pressure, while SME changed only slightly. However, the low-feed, high-temperature condition produced the highest (up to 2.37) anisotropy index (AI; a measure of directional fiber alignment), supporting the hypothesis that stronger thermal treatment and longer residence time improved protein alignment and fibrous structure. Low-temperature conditions produced the firmest texture (hardness up to 183.72 N), showing that firmer products were not necessarily more fibrous. Visual fibrosity was higher in STD (7.35-8.11) than SP (4.02-6.75). These results supported the formulation hypothesis that STD showed stronger hydration and higher visual fibrosity, but AI did not differ significantly between STD and SP under matched processing conditions, indicating that processing conditions, particularly low-feed and high-temperature, were the main driver of structural anisotropy. Fiberlyzer, an image-based method that quantifies fibrousness from elongated structures in macro images, showed a strong positive correlation with AI (r = 0.93), and an expert-guided deep-learning model based on a modified ResNet-18 showed strong agreement with expert visual fibrosity scores (r = 0.89). The third and final part of this study explored extrusion-based 3D printing of plant-based meat using PPI and SPI with different levels of methylcellulose (MC) and pea fiber (PF). SPI (cold swelling) hydrated faster than PPI (non-cold swelling) and produced plant protein ink with higher viscosity. SPI formulations were firmer, reaching hardness up to 23.47 N, likely because faster hydration and higher viscosity formed a stronger, more cohesive printed network. MC showed very high WAC (9.93 g/g) and increased viscosity, which helped to print filaments that set quickly after deposition. However, at higher MC levels, this strong thickening made the ink overly viscous and elastic, limiting relaxation after extrusion and reducing dimensional stability, reflected by more negative dimensional distortion ratio (DDR) values (as low as -0.27), where DDR compares the measured printed dimensions with the intended design, and values closer to zero indicate better shape fidelity. In contrast, PF improved shape retention by providing more gradual water binding and physical support, reducing deformation and bringing DDR closer to 0. This research showed that plant-based meat quality depends on the combination of plant protein functionality and processing. Across TVP, HMMA and 3D-printing products, differences in hydration, viscosity, moisture, temperature, screw speed and formulation shaped processing, structure, and texture, providing an improved theoretical and scientific framework and practical guidance for designing more consistent products.